Abstract: A speech synthesis system with a linear predictive filter. The linear predictive filter utilizes coded reflection coefficients to produce digital signals representative of human speech. A variable interpolation circuit within the linear predictive filter allows a variable number of interpolation steps to be calculated between successive values of reflection coefficients. Additionally, a user programmable option allows the user to select a linear, nonlinear, or a combination form of interpolation based on stored scale data.

Abstract: Operation of a continuous speech recognition system operable according to the dynamic programming technique, is controlled by a state transition diagram in compliance with which word sequences to be recognized by the system with reference to a predetermined number of reference words B.sup.n 's are pronounced. The system comprises a state transition table accessed by the reference words B.sup.n 's to successively produce particular states y's in the diagram and previous states z's for each particular state y. In cooperation with a recurrence value and an optimum parameter table, a matching unit determines a recurrence value T.sub.y (m) and an optimum parameter set ZUN.sub.y (m) according to: ##EQU1## where u and m represent an end and a start point of a fragmentary pattern A(u, m) of an input pattern A representative of a word sequence and D(u, m, n), a similarity measure between the fragmentary pattern A(u, m) and a reference word B.sup.

Abstract: Speech parameter signals may be coded to consist of fewer bits for data reduction in memory. In this system both coded and uncoded signals are stored with an associated control signal which controls selection of corresponding direct or decoder circuit coupling of memory output to a speech synthesizer.

Abstract: A transmitter-receiver synchronizer is described for use in terminating T1 lines from a subscriber line switch at a central time division multiplexed switching system. A shared rate converter provides bidirectional data rate conversion between the 1.544 Mb/s T1 line and the 2.048 Mb/s switching system. Two buffer memories are used, each writing in at one rate, and reading out at another rate.

Abstract: A multichannel digital speech synthesizer comprises a pulse generator storing periodic and aperiodic excitation signals to be processed in a lattice filter according to weighting parameters, such as gain and reflection coefficients, transmitted from a computer via a control unit and a plurality of input modules assigned to respective output channels. Each input module includes a resettable counter for timing the emissions of periodic or aperiodic excitation signals, to generate a voiced or an unvoiced speech element, and for requesting a new set of parameters from the computer upon detecting the end of a validity interval for a current set of parameters; the module further comprises a pair of buffer memories alternating in reading and writing operations under the control of the counter to ensure a continuous flow of parameter sets to the filter.

Abstract: Adaptive differential PCM system using both adaptive quantization and adaptive prediction. The adaptive prediction is carried out in adjacent frequency bands on as many signals as are obtained by weighting the quantized signal by coefficients obtained by spectral analysis in the bands of the quantized signal.

Abstract: In this speech recognition system the array formed by a timewise sequence of speech signal feature vectors includes digital data at each time slot representing both presence/absence and consistency of occurence.

Abstract: This delta modulation system uses the differences between three successive samples of a waveform (the i-1, i, and i+1) such that the number of bits per digitization is reduced from four to two. The value vi of the ith sample is calculated by the equation v.sub.i =v.sub.i -1+f(.DELTA..sub.i-1, .DELTA..sub.i), and f is a special function defined by 16 values stored in a read-only memory.

Abstract: Disclosed is a circuit for determining whether an incoming signal is a speech signal or a data signal. In the preferred embodiment, means for separating the incoming signal into a low frequency component and a high frequency component are provided. The low frequency component consists of frequencies below a specified frequency, namely 1000 Hz. and the high frequency component consists of frequencies above 1000 Hz. Means are also provided for the comparison of the energy level of the high frequency components with a multiple of the energy level of the low frequency components. Also, means are provided for comparing a multiple of the energy level of the low frequency components with a maximum threshold and for comparing the energy level of the high frequency components with a minimum threshold. Further, means are provided for detecting whether the incoming signal has a slope.

Abstract: A method and system for reducing the actual bandwidth required for the transmission of signals which are inherently time and frequency division multiplexed, e.g.

Abstract: A system for use with equipment designed for transmitting a relatively wideband signal over two or more relatively narrowband circuits so as to provide good overall system reliability even though the relability of the individual narrowband circuits is poor.

Abstract: A pulse width modulated digital-to-analog converter for utilization in low voltage, integrated circuit speech synthesis circuitry. A digitally programmable shift register is utilized to generate a pulse, which is generally related to the magnitude of a digital signal. A programmable delay circuit provides finer resolution by converting the least significant digital bits of data into pulse width information of shorter duration than the minimum pulse width generated by the controllable shift register. Pulse width information generated by the shift register and delay circuit is applied to the bases of two crossconnected transistors to drive a speaker or voice coil.

Abstract: Disclosed is a system whereby in a transmit mode analog speech is sampled d converted digitally to 12 bits of accuracy and then fed into a fast Fourier transform (FFT) processor which analyzes the speech into spectral and pitch parameters. These parameters are then quantized into a data stream which acts as an input to a differential phase shift keying modulator. The modulator constructs a multi-tone modem output signal from a 25 tone stack which is converted to an analog signal which is fed into a communications channel. In the receive mode a modem input signal from the communications channel is fed through the same analog to digital converter which was used for input speech. A data stream is now provided which is coupled back into the same digital FFT processor which now operates to provide pitch spectral coefficients which are then separated and used by a synthesizer to reconstruct a speech waveform.

Abstract: Data converter for a speech synthesizer system wherein encoded formant parameters as stored in a memory are decoded and transformed or converted to reflection coefficients in real time by means of a circuit implementing a Taylor series type approximation. The reflection coefficients are then quantized and input to a speech synthesizer which utilizes quantized reflection coefficients to synthesize speech. The use of the coded formant frequency speech data which inherently contains more speech intelligence than reflection coefficient speech data enables a speech synthesizer system which utilizes quantized reflection coefficients to operate at a significantly lower bit rate than would otherwise be possible where reflection coefficients are employed as the speech data stored in the memory.

Abstract: A speech synthesis circuit is provided with a variable frame length data converter and the speech synthesizer is preferably integrated on an integrated circuit chips. The variable frame length data converter reduces the amount of data required to synthesizer human speech at a given quality level. Preferably, a full frame of data includes, a pitch parameter, an energy parameter, a repeat bit and a plurality of speech coefficients. Each parameter or coefficient has a preselected length, but each frame has a variable number of parameters or coefficients associated therewith. The parameters and coefficients are encoded and a particular code of the pitch parameter indicates that the speech is to be unvoiced. An unvoiced frame includes fewer coefficients that a voiced frame and the converter detects this particular pitch parameter and automatically sets the unsent coefficients to zero.

Abstract: A digital speech interpolation system is combined with an adaptive differential PCM (ADPCM), employing a speech detector for detecting speech signals and for discriminating voiced and unvoiced sounds. An adaptive quantization bit assignment to the speech is adopted to cope with any freeze-out condition. And further PCM speech signals with 8 KHz sampling are applied to ADPCM after shifted 250 Hz down and then converted into 6 KHz sampling frequency, thereby attaining a total gain of about 7 without degrading speech quality.

Abstract: There is disclosed a decimator structure which incorporates the cascade of an FIR filter with a low pass recursive filter. The input to the decimator is obtained from a high rate analog to digital converter. The output from the decimator is a low rate digital signal having an increased word length. The decimator serves to reduce the word rate and increase the word length of the output digital signal of the analog to digital converter. In this manner, the low rate digital signal at the output of the decimator can be easily accommodated by the telephone system. The decimator described does not require conventional hardware multipliers and hence, affords efficient and economical circuit components which can be implemented by conventional integrated circuit techniques. A register further reduces the pulse rate for transmission.

Abstract: A device for providing real time speech compression and expansion and method therefor which enables speech data, sampled at a predetermined rate, and consisting of a preselected number of bits per sample, to be compressed in real time by a factor of approximately 2:1, and upon decompression, to be essentially indistinguishable from the original digitized speech. In particular, the apparatus employs a method which involves the digitizing of the input data, determining the maximum amplitude of the data over a short given period, compressing the data by dividing the largest amplitude constant from the data over the given period to obtain a data sample, and storing the reduced data sample and corresponding amplitude constant or gain factor in a suitable storage medium. In the preferred embodiment, the storage medium is a rotating memory or a recirculating shift register.

Abstract: In order to prevent errors and instability which may occur in a speech analysis and synthesis apparatus when the normalized predictive residual power falls to low levels, for example in high-pitched speech, the calculation of linear predictor coefficients from the autocorrelation coefficients of the speech sound is stopped when the normalized predictive residual power falls below a predetermined threshhold level. Either a variable stage synthesis filter is used having its number of stages determined by the number of linear predictor coefficients actually calculated, or a fixed number of stages can be used and a zero value filter stage coefficient supplied to those stages in excess of the number of coefficients calculated. Degradation of speech quality due to quantization and transmission errors can be alleviated by computing the normalized predictive residual power on the synthesis side from the transmitted predictor coefficients and using it to excite the input to the synthesis filter.

Abstract: A voice synthesizer that is responsive to sequences of digital input command words to phonetically synthesize human speech. The system includes control circuits that are responsive to the input command words to introduce an articulated silent phoneme into the speech pattern, vary the duration of each phoneme produced, as well as to vary the overall rate and volume of the speech generated. In addition, the design utilizes inflection assignment for individual phonemes and also employs a glottal waveform which is more representative of human glottis action. The invention also incorporates resonant suppression into the vocal tract to simulate the dampening effect due to the opening of the glottis, and provides closer simulation of human energy content at higher frequencies to improve the quality of the speech generated.